Michael Helle1, Fabian Wenzel1, Kim van de Ven2, and Peter Boernert1
1Philips Research, Hamburg, Germany, 2Philips Healthcare, Best, Netherlands
Synopsis
This
study presents an advanced fully automated approach based on vessel
detection and analysis. It is completely integrated in the scanner console and
allows labeling of the major brain feeding vessels in an efficient and robust
way. Average processing time to find optimal labeling positions for all major brain
feeding arteries is <15 seconds.
Introduction
Super-Selective
Arterial Spin Labeling (ASL) proved to be an efficient method for selective
labeling of individual blood vessels (1). It can be used in conjunction with
different acquisition modules, thereby, generating angiograms of the cerebral
vasculature (2) or perfusion territory maps of the brain (3) and has been
applied in various patient studies (4, 5). However, careful planning of the
labeling spot is required and might add extra time to the imaging protocol.
Recently, first attempts to facilitate the positioning of the labeling spot
have been introduced and compared to images of manually planned labeling spots
(6, 7). This study presents an advanced fully automated approach based on vessel
detection and analysis. It is completely integrated in the scanner console and
allows labeling of the major brain feeding vessels in an efficient and robust
way.
Methods
Measurements
were performed in six healthy volunteers on a 1.5T Ingenia Scanner (Philips,
Best, The Netherlands) using a 16-element head-coil. Automatic planning of the
labeling spots was performed on the basis of a time-of-flight (TOF) scan to
visualize the vascular anatomy of the neck (FOV 200x200x96mm3, voxel
size 1.5x1.5x1.5mm3, 3D fast-field echo acquisition, FA 18°,TR/TE
23/2.3ms, 1:07min scan time). The automated planning consists of a sequence of
image processing steps, including vessel tracking functionality from a
commercial software package (6, 8). Parameters with impact on the final
labeling spot of a particular vessel are based on three rejection criteria:
First, the angular deviation from the z-direction of the magnet. Second, the
extent of the labeling plane into which the course of a vessel should not
re-enter after labeling. Third, the maximum length of a vessel part inside the
labeling plane. The present implementation allows a maximum angular deviation
of 5° in order to reduce possible effects of gradient inhomogeneities to the
labeling efficiency. A width of the labeling plane has been set to w=18mm,
approximating its true width according to the applied labeling gradient in
z-direction. A maximum length of a vessel part inside the labeling plane has
been set to l=22mm, therefore excluding labeling positions of highly curved,
and unsuitably elongated vessel parts. More parameter combinations (w=18-25mm;
l=22-27mm) and their impact on labeling locations were investigated on 10
retrospective TOF acquisitions. Image acquisition for flow territory mapping
was performed based on the automatic positioning of the labeling spots onto
both internal carotid arteries (ICA) and both vertebral arteries (VA). Scan
parameters for super-selective ASL were: 2-pulse background suppression,
segmented 3D GraSE read-out (FOV 240x240x112mm3, voxel size
3.75x3.75x8mm3, FA 90°, TSE/EPI factor 17/15, TR/TE 3913/15ms,
3 averages, labeling duration 1.8s, post-labeling delay 1.8s; 1:57min scan time
per vessel). Subsequently, the labeling
efficiency was calculated by normalizing the signal intensities of flow
territory images with respect to the signal intensity of an additionally
performed non-selective ASL scan.
Results
Figure
1 presents the angiogram and analyzed vessel architecture with calculated
positions of the labeling spots based on two different tracking parameter
settings for one single volunteer. Labeling was effective for each parameter
setting with no significant change in image quality, even though the labeling
spots for the right ICA and VA were in a different position. Visual inspection
of results from altered tracking parameters has confirmed suitability of this
choice and has led to more configurations with no valid labeling focus
otherwise. Average processing time in each volunteer to find optimal labeling
positions for all major brain feeding arteries was approximately 12s.
Successful flow territory mapping was performed (figure 2) and similar labeling
efficiency was achieved in the ICAs when compared to a non-selective ASL scan
(figure 3). However, labeling efficiency decreased in VAs possibly due to
mixing of the blood in the posterior circulation. In three volunteers, only the
signal of one VA was detected as the contralateral VA presented tiny with small
caliber and probably only has minor contribution to the perfusion of the
posterior circulation (figure 1a, 2).
Discussion
Fully
automatic positioning of the labeling spot can be helpful in clinical routine
scan protocols to overcome user-dependent and time-consuming planning. The
presented approach demonstrates fast and reliable planning; slight changes of
the tracking parameters may shift the positioning of the labeling spot in some
vessels, but the differences seem too small to have a significant impact on
image quality. However, this might be different in patients with altered
vasculatures and would require separate investigation. The planning algorithm
presented stable even in small caliber arteries like some of the VAs, however,
as the amount of labeled blood can be too small to generate detectable signal
in the brain, instead, one may consider using elliptical labeling spots to
label both VAs at the same time (9).Acknowledgements
No acknowledgement found.References
-
Helle M, Norris DG, Rufer S, Alfke K, Jansen O, van
Osch MJP. Superselective pseudocontinuous arterial spin labeling. Magn Reson Med
2010;64:777-786.
- Jensen-Kondering U, Lindner T, van Osch MJ, Rohr A,
Jansen O, Helle M. Superselective pseudo-continuous arterial spin labeling
angiography. Eur J Radiol 2015;84:1758-67.
- Hartkamp NS, Helle M, Chappell MA, Okell TW, Hendrikse
J, Bokkers RP, van Osch MJ. Validation of planning-free vessel-encoded
pseudo-continuous arterial spin labeling MR imaging as territorial-ASL strategy
by comparison to super-selective p-CASL MRI. Magn Reson Med 2014;71:2059-70.
- Helle M, Rüfer S, van Osch MJ, Nabavi A, Alfke K,
Norris DG, Jansen O. Superselective arterial spin labeling applied for flow
territory mapping in various cerebrovascular diseases. J Magn Reson Imaging.
2013;38:496-503.
- Richter V, Helle M, van Osch MJP, Lindner T, Gersing AS, Tsantilas P, Eckstein HH, Preibisch C, Zimmer C. MR
Imaging of Individual Perfusion Reorganization Using Superselective
Pseudocontinuous Arterial Spin-Labeling in Patients with Complex
Extracranial Steno-Occlusive Disease. AJNR Am J Neuroradiol. 2017;38:703-711.
- Helle M, van de Ven K, Wenzel F. Automatic Planning
for fast and robust Flow Territory Mapping. Proc. Int. Soc. Magn. Reson. Med.
2017:3625.
- Lindner T, Jansen O, Helle M. Hough-transform based
detection of vascular structures applied to automate and accelerate planning of
super-selective Arterial Spin Labeling. Proc. Int. Soc. Magn. Reson. Med.
2017:3823.
-
Bescós J, Sonnemans J, Habets R, Peters J, van den
Bosch H, Leiner T. Vessel Explorer: A tool for quantitative measurements in CT
and MR angiography. Medicamundi 2009:53/3.
- Helle M, Rüfer S, van Osch MJ, Jansen O, Norris DG. Selective
multivessel labeling approach for perfusion territory imaging in
pseudo-continuous arterial spin labeling. Magn Reson Med 2012;68:214-9.